Summary 1. Fishes can often rebound numerically and distributionally from short-term (i.e. seasonal) drought, yet their capacity to recover from decades or centuries of drought is less apparent. An exceedingly warm and dry period swept the intermontane west of North America ca. 7500 years BP, concomitant with an abrupt extinction of >35 mammal species. Were larger fishes in mainstem rivers also impacted by this drought?
2. The Colorado River Basin encompasses seven states in western North America and drains 600 000 km2. Its endemic mainstem fish community is ancient (i.e. Miocene) but depauperate.
3. We evaluated one widely distributed candidate species (flannelmouth sucker, Catostomus latipinnis) for basin-wide genetic and geographic structure at three fast-evolving mitochondrial (mt) DNA genes, ND2 with 589 bp and ATPase 8 and 6 with 642 bp. It is hypothesized that a concomitant signature would be present in the mtDNA of this species, if indeed it had been seriously bottlenecked by post-Pleistocene drought. A total of 352 individuals were sequenced from 24 populations (4–40 individuals/population; average of 14.7).
4. Only 49 unique haplotypes were found, 53% of which represented single individuals. Haplotype diversity was high (0.905 ± 0.007) whereas nucleotide diversity was low (0.002 ± 0.000).
5. A significant and positive geographical cline (P < 0.001) in nucleotide diversity was observed as sampling locations progressed upstream from southwest to northeast. These results divided the Colorado River Basin into three reaches: the lower reach with six populations and 83 individuals; the upper reach with seven populations and 83 individuals; and the middle reach with 11 populations and 186 individuals. An analysis of molecular variance (amova) revealed that 81.5% of the total genetic variation was within populations, 16% among populations within reaches and 2.5% among reaches. Only the last was significant. Populations from the three reaches diverged from one another by 3400–11 000 years BP. Haplotype distribution suggested populations in the upper Colorado River are expanding.
6. The lack of genetic variation and recent coalescence of lineages in C. latipinnis are unusual given its fossil history, broad geographical sampling, the rapid rate of mtDNA evolution and the number (and evolutionary rate) of the genes examined. The most parsimonious explanation for these data is a rapid expansion following a recent period of low effective population size at the end of the Pleistocene.
7. The intense drought is suggested at the end of the Pleistocene (late-to-mid-Holocene), severely impacted not only large mammals but also larger fishes in western North American rivers. These perspectives have important implications for management of endangered and threatened species in this region.